gdbserver: fix killed-outside.exp

killed-outside.exp regresses with "maint set target-non-stop on".  The
logs show:

 (gdb) continue
 Continuing.
 infrun: clear_proceed_status_thread (Thread 9028.9028)
 infrun: proceed (addr=0xffffffffffffffff, signal=GDB_SIGNAL_DEFAULT)
 infrun: proceed: resuming Thread 9028.9028
 Sending packet: $Z0,3615a03966,1#4b...  Notification received: Stop:X9;process:2344
 Packet received: E01
 Sending packet: $Z0,3615a13970,1#47...Packet received: E01
 Sending packet: $Z0,3615a14891,1#4a...Packet received: E01
 infrun: resume (step=0, signal=GDB_SIGNAL_0), trap_expected=0, current thread [Thread 9028.9028] at 0x4005e4
 Sending packet: $vCont;c:p2344.2344#1a...Packet received: E.target not running.
 Sending packet: $qXfer:threads:read::0,fff#03...Packet received: l<threads>\n</threads>\n
 Sending packet: $vStopped#55...Packet received: OK
 Unexpected vCont reply in non-stop mode: E.target not running.
 (gdb) remote_async_inferior_event_handler
 infrun: target_wait (-1.0.0, status) =
 infrun:   9028.0.0 [process 9028],
 infrun:   status->kind = signalled, signal = GDB_SIGNAL_KILL
 infrun: TARGET_WAITKIND_SIGNALLED

 Program terminated with signal SIGKILL, Killed.
 The program no longer exists.
 infrun: stop_waiting
 infrun: clear_step_over_info
 infrun: stop_all_threads
 remote_thread_exit_events(1)

Note the "Unexpected vCont reply" error.

I traced it to a problem in status_pending_p_callback.  It resumes an
LWP when it shouldn't.

gdb/gdbserver/ChangeLog:
2015-11-30  Pedro Alves  <palves@redhat.com>

	* linux-low.c (thread_still_has_status_pending_p): Don't check
	vCont;t here.
	(lwp_resumed): New function.
	(status_pending_p_callback): Return early if the LWP is not
	supposed to be resumed.
This commit is contained in:
Pedro Alves
2015-11-30 16:05:23 +00:00
parent 65706a29ba
commit a681f9c913
2 changed files with 31 additions and 6 deletions

View File

@ -1,3 +1,11 @@
2015-11-30 Pedro Alves <palves@redhat.com>
* linux-low.c (thread_still_has_status_pending_p): Don't check
vCont;t here.
(lwp_resumed): New function.
(status_pending_p_callback): Return early if the LWP is not
supposed to be resumed.
2015-11-30 Pedro Alves <palves@redhat.com>
* linux-low.c (handle_extended_wait): Assert that the LWP's

View File

@ -1525,12 +1525,6 @@ thread_still_has_status_pending_p (struct thread_info *thread)
if (!lp->status_pending_p)
return 0;
/* If we got a `vCont;t', but we haven't reported a stop yet, do
report any status pending the LWP may have. */
if (thread->last_resume_kind == resume_stop
&& thread->last_status.kind != TARGET_WAITKIND_IGNORE)
return 0;
if (thread->last_resume_kind != resume_stop
&& (lp->stop_reason == TARGET_STOPPED_BY_SW_BREAKPOINT
|| lp->stop_reason == TARGET_STOPPED_BY_HW_BREAKPOINT))
@ -1587,6 +1581,26 @@ thread_still_has_status_pending_p (struct thread_info *thread)
return 1;
}
/* Returns true if LWP is resumed from the client's perspective. */
static int
lwp_resumed (struct lwp_info *lwp)
{
struct thread_info *thread = get_lwp_thread (lwp);
if (thread->last_resume_kind != resume_stop)
return 1;
/* Did gdb send us a `vCont;t', but we haven't reported the
corresponding stop to gdb yet? If so, the thread is still
resumed/running from gdb's perspective. */
if (thread->last_resume_kind == resume_stop
&& thread->last_status.kind == TARGET_WAITKIND_IGNORE)
return 1;
return 0;
}
/* Return 1 if this lwp has an interesting status pending. */
static int
status_pending_p_callback (struct inferior_list_entry *entry, void *arg)
@ -1600,6 +1614,9 @@ status_pending_p_callback (struct inferior_list_entry *entry, void *arg)
if (!ptid_match (ptid_of (thread), ptid))
return 0;
if (!lwp_resumed (lp))
return 0;
if (lp->status_pending_p
&& !thread_still_has_status_pending_p (thread))
{